Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Civil and Environmental Engineering


Dr. Craig Miller


Time histories of wind speed and direction from 394 surface observation stations were obtained to calculate synoptic 50-year return period wind speeds for 11 countries in Europe. Preliminary investigation indicated wind speed differences along national borders were successfully reduced by application of a simple consistent methodology to wind speed data. This study considers the ideal methodology for calculating synoptic 50-year return period wind speeds.

Wind speed data requires standardisation through quality control measures, exposure correction and adjustment for disjunct sampling. A quality control algorithm was successfully applied to identify shifts of monthly mean wind speeds and data conversion issues. Three exposure correction models were evaluated and two-layer models were found to perform better than internal boundary layer models. The differences arise as a result of how the models adapt to an upstream change of roughness. Furthermore, an empirical model was formed to correct observations at stations which were not recording measurements hourly.

Extreme value analyses were carried out using a robust estimator to fit the extreme value distribution type I to storm and yearly maxima. The latter was found to provide more consistent results. Comparison of the resulting 50-year return period wind speeds to existing literature found that several regions were in good agreement, while other regions exhibited similar spatial variation but greater magnitudes. The differences in magnitude were partially related to exposure correction methods, thus lending support to the importance of a single consistent methodology. Directional factors were calculated and subsequently grouped into six regions exhibiting similar directional characteristics.

Background wind fields were calculated from mean sea-level pressure data using the geostrophic approximation and consideration of other improved approximations, however, variations in the pressure field led to a breakdown of the methodology. A background 50-year return period wind field calculated from upper-level wind fields was significantly lower than surface wind speed estimates due to spatial and temporal smoothing. Finally, assimilation of the 50-year return period wind speeds from surface observations and the background wind field was explored using the Bratseth scheme for statistical interpolation. The Bratseth scheme provided an overall 50-year return period wind speed map.